Battle of the retro Unix desktops: NsCDE versus CDE

If the real CDE is too much hard work for you or for your computer, there's a new version of the Not So Common Desktop Environment. Very nearly a decade ago, we reported that the official Common Desktop Environment had been made open source. In its day, CDE was pretty much the unified desktop environment for commercial Unix …

[Golgafrinch]

You can not be serious, man ...

I can't believe someone actually wrote "...before the relative sophistication that Windows 95 delivered ..."

Remember the WPS that came with OS/2 2.0? (Alright, we subsequently had to live with CDE inspired disimprovements which no one had asked for.)

[Jan 0]

Re: You can not be serious, man ...

By "relative sophistication" I assumed that was comparing it with DOS.

[Liam Proven]

Re: You can not be serious, man ...

[Author here]

Compared to Windows 3.1 -- yes, very much so!

That's why about 18 of the leading 20 Linux desktops now are (mostly fairly poor) re-implementations of Windows 95, whereas there are a grand total of, er, _no_ re-implementations of WPS.

I wish there were. I bough OS/2 2.0 with my own money and used it for years. I quite liked WPS; it was state of the art in its time.

[chasil]

HP VUE

You might find it interesting to know that the predecessor to CDE was HP VUE under HP-UX.

I graduated from the university of Iowa. When I matriculated, the engineering computer lab had Macs and Apollo Domain workstations. The Apollos migrated to PA-RISC running HP-UX v9.

This was shortly before the attempt on DCE standardization, and I think that HP submitted VUE for this reason.

https://en.m.wikipedia.org/wiki/Visual_User_Environment

One unsung jewel of CDE was the dtksh, a Korn shell variant that allowed direct access to Xlib, Xt, and Motif widgets from a shell script. This code was owned by Novell, and the author documented the final version in this book:

https://www.amazon.com.au/Desktop-Kornshell-Graphical-Programming-Pendergrast/dp/0201633752

Good times.

[GuldenNL]

Re: HP VUE

You’re right! I forgot all about it while reading about OS/2

[Golgafrinch]

Re: HP VUE

@chasil:

You're right, the dtksh was a gem (no pun intended). Unfortunately, most sysadmins I got to know resolutely stuck to ksh88, and never even graduated to ksh93.

[Jim 59]

Re: HP VUE

Loved those Apollo Domain systems, including the Window Manager (wm) shell at the bottom of the screen, where you could issue direct commands to the window system, similar to dtksh, perhaps.

HP VUE was also nice. A bit basic, even at the time, but solid.

[therobyouknow]

Nortel/BNR

I remember using these at Nortel/BNR in the mid 90s.

[unimaginative]

Re: You can not be serious, man ...

"That's why about 18 of the leading 20 Linux desktops now are (mostly fairly poor) re-implementations of Windows 95, whereas there are a grand total of, er, _no_ re-implementations of WPS."

Not true. A lot of desktop environments are very flexible and may default to looking like Windows 95 because its simple and familiar to new users but this is very superficial - but they are far more flexible and can look and feel completely different. None of the DEs I have used in the last few years (KDE, XFCE and Enlightenment) need to be anything like Windows 95, and I find them better with a different configuration. I do not think Gnome even looks like Windows 95 by default in most cases, does it?

I think the problem may be is that making DEs look superficially like Windows (taskbar, start button, in a single panel etc.) creates false expectations which lead to a poor experience when they do not work like Windows. Maybe distros should use more adventurous defaults?

[Mac Logo]

Re: You can not be serious, man ...

> I do not think Gnome even looks like Windows 95 by default in most cases, does it?

Baby and Bathwater springs to mind - often seems like the Gnome-devs looked at what did work in Win95 and threw those out first.

[martinusher]

Re: You can not be serious, man ...

The only thing that Win95 delivered was -- finally -- built in Internet support.

Anyway, in 'ix' systems a desktop is an application, just like any other. You can run one or another and even run it remotely over X-Windows if you want. Its only the Windows mindset that sees it as an intrinsic part of the operating environment.

[AJ MacLeod]

Re: You can not be serious, man ...

My Windows 95 didn't have built in Internet support. (It also became so unreliable that it drove me to Linux and within two years it was banished for good.)

I used CDE quite a bit after that on SUNs but it all (NT, Win98, CDE) felt very old fashioned and clunky compared to the desktops I was by then used to on Linux; KDE or Enlightenment especially.

In the end I settled on WindowMaker and am still a very happy user.

[John Brown (no body)]

Re: You can not be serious, man ...

"Anyway, in 'ix' systems"

FWIW, the usual abbreviation is *nix systems. Ix Systems is a US company founded on open source and building commercial stuff, which I what I immediately think of when I see the fairly rare use of "ix systems" as an abbreviation if Unix-like.

[gotes]

Re: You can not be serious, man ...

What does "built in internet support" mean? I remember using TCP/IP on WFW 3.11, but it didn't come with a bundled web browser. I was able to use the internet without installing 3rd party network drivers.

[martinusher]

Re: You can not be serious, man ...

If I recall correctly you had to add a service pack or third party code to get a TCP/IP stack.

[Agamemnon]

Re: You can not be serious, man ...

Good ol' TrumpetWinsock.

[Timochka]

Pff.

Open Look or I'm not interested.

[Golgafrinch]

The main appeal of openlook was that it was dead easy to create your own menus.

As for icon handling - yes, that worked great on my Sun4/110 with a 19" monitor, but less so on a poxy Linux PC, where it turned out to be a complete waste of screen estate (as, BTW, all 'modern' window managers are - when I had my last aesthetic and ergonomic clash, I sang "I really don't like it, fuck the task bar".)

[Dr_N]

Exactly. .

OpenWindows with olvwm for multiple (but connected) workspaces.

Never been equalled.

[VoiceOfTruth]

RAM usage

-> CDE used a whopping 892 MiB of RAM, over three times as much as NsCDE.

This I do not understand. Back some time around Y2K, I ran CDE on Solaris 7 on intel. The machine had 64MB of RAM. I really don't understand why modern OSs seem to need 10 times (or more) memory compared to running the same thing from 20 years ago. Where does all this memory go?

[Golgafrinch]

Re: RAM usage

"Where does all this memory go?"

It went agile.

[Swarthy]

Re: Where does all this memory go?

layers and layers of shims and emulation, would be my guess.

[m4r35n357]

Re: RAM usage

I suspect a huge chunk (but not all!) of it goes with high resolution screen support. 640x480 with 8-bit colour was fairly typical at the time.

[Anonymous Coward]

Re: RAM usage

Well... a 8k monitor is 7680 * 4320 * 32bit = 126Mb.... so only 500Mb more to explain away!

[Richard 12]

Re: RAM usage

Times 2 for front and back buffer, so that's actually 252Mb at minimum.

Plus the front and back buffers for every individual window to feed the compositor.

It adds up surprisingly quickly, and all of this is just to prevent flicker and hall-of-mirrors effects.

I'm sure you remember how a crashed or frozen application used to be able to take over the entire screen...

[VoiceOfTruth]

Re: RAM usage

I had 1600x1200 back then. I can't remember if it was 65k colours or 16 million. But it wasn't 256 colours.

[Flocke Kroes]

Re: RAM usage

For early '90s my monitor could do 1600x1200 but my computers could only do two colours at that resolution. I had to drop to 640x480 (and wait a few seconds for resynchronisation) to get 256 colours. It took years and several house moves with that 50kg monitor before I could use 1600x1200 with 2^24 colours. You must have had some top notch kit back then.

[Sandtitz]

Re: RAM usage

"For early '90s my monitor could do 1600x1200 but my computers could only do two colours at that resolution."

That was due to lack of graphics memory on you adapter. 640x480x256 (8-bit) is just 300kB, so perhaps you had only 512kB on your card? A low-cost Trident 8900 ISA card in my 286 PC had 1MB onboard RAM so I could use 1024x768x256 graphics.

Anyway, OP wrote "some time around Y2K", not early 90s.

My 22" Samsung 1200nf from 2001 (?) had 2048x1536@75Hz resolution (almost unusable); Sony GDM-W900 and (later on) FW900 were the ones everyone drooled after but they were ludicrously expensive.

2048x1536 with 32-bit color depth equals 12MB, so my late 90's era 16GB Nvidia TNT card could output it, but the limitations of analog output and the less than perfect geometry of that CRT with only 75Hz produced a picture that wasn't satisfying. (also, any 3D gaming like Half-Life wasn't anywhere near playable at that resolution)

[Zippy´s Sausage Factory]

Re: RAM usage

Trident 8900 cards bring back memories of doing corporate PC support back in the early 90s. I used to like Tridents - for some reason they tended to be easier to set up than the other horrors clients used to waste money upgrading to (because WordPerfect on DOS required the absolute latest graphics hardware, naturally... or perhaps that was the six disks with "flight sim" written on in biro that were sat in the boss's desk...)

[Missing Semicolon]

Re: RAM usage

Oh, yes, the Trident TVGA card.

Would do 132 columns by 50 lines (text mode) on a monochrome 14 inch monitor. The dogs, if you were young, and had eyeballs that still worked....

[VoiceOfTruth]

Re: RAM usage

It wasn't really top notch kit. By the time Y2k came around, and even before then, 1600x1200 was within reach of mere mortals who wanted to spend a bit but not break the bank.

[Peter Gathercole]

Re: RAM usage @m4r35n357

Don't know what you were doing back in the mid-90s when CDE was the default. but I would say that the minimum for a UNIX workstation back in the day was 1024x768 at 8 bit colour, but the majority of the systems I was using at the time had 1280x1024 with at least 16 bit colour visuals.

The 640x480x8 may have been a common resolution for PCs, but not for UNIX workstations.

On the subject of where the memory went, it was not normal to have shared memory for the processor and graphics adapter. Anything running on a true workstation would probably have it's own display adapter memory, which is often what limited the resolution/colour depth. So the lost memory was probably not down to the screen.

For the version using 800+MB, I would suggest that the author should look at the options used to compile the source. I'll bet that the binaries are not stripped, the diags. are all turned on by default, and the optimizer was not running aggressively enough.

In the mid '90s, we had CDE running to X-Stations, with about 12 per IBM RS/6000 320H each of which only had 80MB of memory, with some remaining free for other processes. So CDE was really not that heavy.

[John Riddoch]

Re: RAM usage

We had CDE running on Sparcstation 4s & 5s back in the day - I think they only had 16MB or 32MB of RAM. Some of the SS5s only had a 500MB hard drive (that was fun shoe-horning Solaris onto...), let alone memory.

One part might also be a move to 64-bit which has a tendency to inflate binary sizes, but that should only be a doubling at most and probably much less.

[An_Old_Dog]

Re: RAM usage

Perhaps the NsCDE folks put some effort into not having irrelevant dependencies built into their code. I recall being outraged when I discovered Evince, a PDF reader, had a dependency on a GPS-related package.

[ChrisC]

That brings back memories (some almost positive) of my time spent working on HP-UX workstations at uni back in the mid 90's - hateful things with unergonomic mice and a propensity to crash far more often than any other system I had access to at the time, but generally always available to use (presumably because of these traits) which was handy if all the PC labs were fully occupied and you needed to get online to check something in a hurry.

It also makes it clear just how far we've come (fallen) between those days of UI design and what our eyeballs are forced to deal with these days - when you find yourself looking at a screengrab of something intended to replicate a 30 year old UI, and think "my god, just how good does that look!", you know there's something badly wrong with modern UI design...

[Ken G]

Those monitors did keep a room warm though.

[Denarius]

crash ?

Odd, my HPUX workstations stayed up for months. My personal one got worked hard adminning multiple HPUX servers running payroll and development of said payroll application. Eventually bought my own as they were great to code on. Never did get my head around dtksh but liked the idea. CDE in 32 meg or 64 MB RAM worked OK and had a far better display than any PC I saw until early 2000 in an office. Miss it today, but due to HPs inane decision to allow only "HP" disks to be detected I went with AIX 4.3.

[TReko]

UI Fashion

Yep those old user interfaces worked better and looked better.

Now everything is flat. Windows have no borders, buttons look like the rest of the page.

I guess we've gone from functional design made by engineers to UI designers who would otherwise found jobs designing clothes no one wears.

[VoiceOfTruth]

Re: UI Fashion

I am absolutely fed up with the flat look. I wonder why I need a good graphics card if I get something equivalent to twm. I don't play games on work machines.

[therobyouknow]

Nortel/BNR

I remember using these at Nortel/BNR in the mid 90s to build SDH fibre optic equipment embedded software in C/C++.

[Peter Gathercole]

Digital alchemy

"Its terminal emulator can't handle modern apps such as htop or the Tilde text editor".

This is a problem of lack of understanding of terminals and terminal emulators by the system administrator. Similar problems are faced by using Putty or terminals on Linux systems to access proper legacy UNIX systems.

The issue is that not all terminal emulators are vt220, xterm et. al. compatible, and in fact xterm is not a very safe setting for the TERM environment variable used to condition the terminfo entries, as there have been just soooo many mostly compatible, but ultimately not the same versions of "xterm" across the years.

I believe that the correct setting for TERM with the CDE terminal emulator should be "dtterm", but I would suspect that many Linux systems do not have a dtterm terminfo entry, so fall back to xterm, or xterm-256color or something similar. This will almost certainly not match the capabilities of the dtterm terminal emulation.

The common problems are:

Function keys not being recognized correctly

Non-7-bit-ascii characters do not work correctly, especially box draw characters

Any colour support will be very spotty

Some cursor movement operations do not work correctly.

Many of these problems can be fixed at one fell swoop, by identifying the location of the dtterm terminfo file, and making sure there is a copy in the appropriate place for the hosting OS that you are using (unless, like me, you add to the terminfo database with local additions).

The one that will possibly cause a problem is the font that is used, as I'm pretty certain that you will have to have iso8859 fonts in your font path, rather than just UTF-8 ones, unless the version of dtterm on NsCDE has been altered for UTF.

I've just fired up an AIX 5.3 system, and installed an original version of CDE on it, and then run a dtterm via X11 onto a RHEL 8.6 system, and things work pretty much OK, although I don't have "-dt-interface user-medium-r-normal-m" at any size in my default font path.

You cannot imagine how frequently I find I want to take a UNIX or Linux administrator who accepts the wrong characters for box draw or unrecognized function keys as something normal, and try to shake some knowledge into them, because it is nearly always user error, not a problem with the system! UNIX was written to allow *LOTS* of different terminals types to use the system correctly, and Linux mostly inherited the capabilities.

Things have become both more simple, while at the same time more complex with the commodification of UNIX-like OSs, such that what was once well known appears almost like alchemy nowadays.

[Anonymous Coward]

Re: Digital alchemy

I bet you have a *massive* beard o.o

[Peter Gathercole]

Re: Digital alchemy

Nope. Can't stand the things. Too itchy.

But when I've been lazy, and not shaved for a few days, it is quite grey.

[KSM-AZ]

Re: Digital alchemy

No, I do not want f1 to open a help window for the terminal emulator. I want it to perform what the f1 label says in the window. And if you want to emulate a vt100 (vs a wyse 50/60 or an adm3 <grin>) then give me a keyboard without a backspace key, and I want the dip switches underneath. Early VT terminals were total crap (adm3a was cool looking, ctrl-h to backspace). The wyse 60 was sweet, and had a 132 column mode.

I think termcap is better than terminfo, and curses is aptly named.

[Peter Gathercole]

Re: Digital alchemy

Re. termcap vs. terminfo

Termcap you could edit with just an editor. In my view, terminfo is a more comprehensive database, but the problems of compiling the source using tic (and the fact that AT&T UNIXes did not have an 'untic' decompiler shipped as standard) often caused problems.

If you can find it, I recommend finding a version of termcap from before it got cleaned up (BSD 2.x and BSD 3.x tapes), and then read some of the comments about terminals like the "Super Beehive", and some of the Ann Arbour terminals. Whoever created them really didn't like those terminals!

I never really liked Wyse terminals. The Wyse 50 and Wyse 60 had a very clumsy byte-command sequence, almost as bad as the native IBM 3161, 3151 etc. (which were really bad, in that you could not turn off individual attributes, you had to remember the state, turn them all off, and then turn on the ones you still wanted).

DEC VT50 and particularly VT52 terminals were IMHO a good compromise between complexity and function. The command set was quite simple and easy to program, although the terminals themselves were a bit clunky. I found the ANSI 3.64 (e.g. VT100 and later) command set too complicated, and there really wasn't a ANSI 3.64 standard terminal, as the protocol allowed vendor specific private additions. If I had to aim for a 'standard' ANSI compatible terminal, I guess that a VT102 would be the closest, although the lack of very many function keys was a problem (which is why there was an application keypad mode). But the way DEC mapped the VT100 keys to the original VT220 (but fixed in VT420 and later) terminals was just bizarre (PF1-PF4 above the numeric keyboard were F1-F4, the keys that should have been F1-F5 were dedicated hardware (like "Setup", "Break" etc.) there was no F5, F15 was marked "Help" and F16 was marked "Do" and F17-F20 was above PF1-4.

Personally, the best terminal I found was a Falco 5220, which was a good VT220 emulation, with quite comprehensive other modes, plus Tektronik 4014 emulation, and also allowed both serial ports to be used as seperate virtual terminals. Shame they were a bit fragile, though.

[J.G.Harston]

Re: Digital alchemy

Everybody seems to put lots of effort into implementing (and documenting) terminal output, but always seem to completely forget about keyboard input. It took me months of trawlling through the intertubes to actually track down actual documentation of what various non-typewriter keys were supposed to emit and/or actually emitted, and all the inconsistancies and incompatibilities.

Eg:

<esc>[14~ - F4

<esc>[15~ - F5

<esc>[17~ - F6

WTF? What happened to <esc>[16~ ? ? ?

[Peter Gathercole]

Re: Digital alchemy

Um. Document? Try running "od -bc" in a terminal window, and then press the required key a few times.

Also worth knowing how to run xev, which will tell you what X11 key symbol is generated by a key, so you can re-map the key if you want.

For a terminal emulator, the keyboard does not generate the escape sequence. The terminal emulator recognises the key press from it's hardware scan code, and generates the correct key sequence depending on the emulation (similar processes happened in real hardware terminals and emulators running on PCs,)

As many xterm and derived terminal emulators were originally based on vt102 terminals, and then extended with vt220 capabilities (although they generally also included Sun termtool emulations).

The vt220 terminal with the LK201 keyboard was eccentric compared to other terminals, and designed specifically to work with VMS or other DEC operating systems. That is the reason why there is a key marked Do and one marked Help, and the Esc key was not used on those OSs (there is also a orange Gold key and a Blue key). Although they were very popular, using them on UNIX was a bit odd, and most people set them to vt100 mode, just so there was an Esc key.

It's interesting that DEC sold vt220s for VAXes running Ultrix, and it was little better on that than other terminals. Later vt terminals became a bit more configurable. I believe that vt320 terminal allowed you to remap the top left key to escape, and the keyboard on vt420 terminals had the function keys numbered, and allowed the top left keys to be actually used as F1-F5.

Once you get gnome-terminal, kterm, dtterm and many other derivatives of xterm which all deviated from whatever xterm was current at the time, things got very messy.

[Pete 2]

Opening a window on the windows

> If you have a big screen and like a fairly minimal desktop experience that's highly customizable, and you're not fond of taskbars and full-screen app browsers

It would be most enlightening if it was possible to profile all the various desktops to see exactly which features users actually used.

I have a sneaking suspicion that the list would be quite short: open a window, resize it, move it, run an app in it, close it.

Features after that being little more than prospective security holes.

[Flocke Kroes]

Re: Opening a window on the windows

Some people do have a very different work flow: mostly I use keyboard shortcuts to switch between virtual desktops. Desktops are most often a full screen browser or four terminals. As all of these default to the right size and are automatically placed to not overlap I almost never move or resize windows.

I suspect most people only use a fraction of what their environment provides but there may well be plenty of strange people who do not use the most popular subset and would be upset if their favourite features went missing.

[Gene Cash]

Re: Opening a window on the windows

"Window manager" describes a "desktop" for me... I'm trying to track and manage what I'm doing with all those windows. Where's all my EMACS windows? where's all my browser windows? Where's all my xterms? Where is my Thunderbird email window?

How quickly can I switch to the EMACS editing /etc/hosts? Where is my last Google results browser session?

Some of the HUGE things that pisses me off about Windows 10 at work is that A) it takes 6-7 seconds to alt-tab back to a window and 2) when you alt-tab in Edge, it switches between browser tabs, not applications! don't screw with the meaning of a very frequently used key! So Windows fails on both consistency and speed.

Edit: this is obviously why I hate tabbed applications... I want to be able to configure my damned window manager and be done with it... not be at the whim of every asshole with a tabbed window toolkit to decide how he wants it to behave and doesn't make it configurable.

[Richard 12]

Re: Opening a window on the windows

The convention in Windows is that Alt+Tab switches between top-level windows, not applications. That's been the case since 1.0a, possibly earlier. Dunno when the Shift modifier added "cycle backwards".

IIRC, Ctrl+Tab switches between MDI sub-windows. Quite why Microsoft decided that Edge would behave like multiple applications instead of MDI is ... unclear.

Then there's Win+Tab... so many ways.

TBH, these days I have so few windows open that it doesn't matter much. Most of the applications I use start so fast that I only tend to leave file browser windows open, plus outlook and the couple of others that take an age to start.

[david 12]

Re: Opening a window on the windows

Edge Windows are individual processes. They are drawn over the Edge window, but that's just the way they are drawn.

That's the technical reason, but it illuminates a little bit of Browser history. MS wanted the screen to be the browser window, (active desktop, IE part of the OS). And the IE6 tabs, along the bottom of the screen in XP, were all individual applications.

But the users wanted the tabs inside the application window, the way Word and Excel worked in Win95. And Firefox gave them that -- the killer feature of Firefox. So, after loosing the second browser war, MS was forced to come up with a browser that had in-window tabs. But soon after went back to individual processes for individual windows.

FF eventually followed with separate processes for separate windows, but although they've improved, they've never been able to make it work right. The FF processes stillbloat out until they've taken all the available memory. I held out with http and NoScript for as long as I could, but most of the internet now demands js and https, no matter how big the demand on the browser and OS.

[Richard 12]

Re: Opening a window on the windows

No, that's not the reason. It was an intentional choice by Microsoft.

Chrome follows the convention - Ctrl+Tab between tabs in a particular top-level window, Alt+Tab between top-level windows.

The separate (sub) processes are an implementation detail of how Chromium decided to firewall between tabs. Once they get their content hosted in the same top-level window, you get the conventional behaviour for free.

So Microsoft had to make a deliberate effort to use Alt-Tab that way.

[chivo243]

before my time

but unix and a desktop environment? I thought it was cli all the way?

Paris doesn't know either...

I await the schooling to come!

[Liam Proven]

Re: before my time

[Author here]

That started to change, and to end, in the 1970s.

Graphical workstations started appearing in 1980. One of the first pioneering machines was the PERQ:

https://en.wikipedia.org/wiki/PERQ

Its Accent OS evolved into Mach which is the kernel in Apple macOS.

Sun entered the market with the Sun-1 by 1982:

https://en.wikipedia.org/wiki/Sun-1

By the mid-to-late 1980s, most of the proprietary UNIX vendors were offering graphical workstations with GUIs based on the X window system: Sun, Hewlett Packard, IBM, DEC, etc.

Basically one way to look at this is that in the mid-1970s, the microprocessor arrived, which enabled microcomputers, relatively small, inexpensive, single-user (usually single-tasking) machines which were affordable. CP/M became the dominant OS.

When once a department might have a minicomputer with a bunch of dumb terminals on serial lines, now, they got standalone microcomputers... sometimes with the same dumb terminals attached to them, in the early days.

The big minicomputer vendors responded to this by making single-user minicomputers with graphics displays: workstations.

These evolved into desk-side pedestals, and the OSes evolved to use the graphics display for UI not just for showing charts and visualisations.

The advent of RISC processors let workstations shrink down to slimline desktop cases, "pizza box" machines, with a big CRT monitor on top, running proprietary UNIX variants, most using the X window system to draw a GUI.

They were aspirational machines for little teenaged nerds like me, costing as much as a good new car.

What arguably changed that was the Amiga, which made a multitasking multimedia desktop computer affordable enough for the home, and then the Acorn Archimedes, which delivered that with the same level of CPU power as a UNIX workstation that cost £20,000-£30,000 in a home computer.

[Anonymous Coward]

Re: before my time

You could turn that post into an article.

[chivo243]

Re: before my time

That's educational! Thanks for the history lesson, Liam! Much appreciated!

[Peter Gathercole]

Re: before my time

Of course, there was the half-way-house offered by Bell Labs./AT&T with their 5620 and 630 Blit (sometimes called Bell Labs. Intelligent Terminal). Later incarnations like the 730 were more like full blown workstations.

Really interesting multi-windowed intelligent terminals connected to larger UNIX systems, originally down a RS232 serial link using a protocol suite called mpx and later Layers, but later connected via early twisted pair Ethernet, called Starlan by AT&T. I first rememember seeing pictures of 5620 terminals in some AT&T UNIX SIII documentation that was included with the Bell Labs. UNIX edition 7 tape that we had sent from the US. That would have made it around 1982, and I guess that they were knocking around inside Bell Labs before that.

As I was the 'terminal king' at one of my jobs, I had a 630 on my desk.

What made them intelligent was the fact that they could download and execute application code to handle specific functions. I particularly fondly remember CIP, which was an interactive graphics editor that generated PIC graphics files to be included in Troff documents, and a Troff output document preview tool.

There were also games, including a whack-a-mole type game that had pictures of various Microsoft executives that you had to hit, and there was also a version of asteroids, sliding block puzzles, Eyes (like Xeyes), and a couple of mouse-driven terminal emulators and text editors.

[Hero Protagonist]

Re: before my time

Spent a good chunk of my career using those terminals! Loved being able to open a couple of 60 row by 80 column windows on the 5620, which was a tall portrait orientation screen.

Did you have the GEBACA (Get Back At Corporate America) game? Corporate logos flying around the screen that you had to shoot. I think it was the AT&T logo that when you shot it, spawned the logos of the Baby Bell companies that were divested in the 1984 breakup. And then there was Crazy Eddie, who spewed out a stream of words like his TV commercials, and you had to shoot the individual words. Good times!

[Peter Gathercole]

Re: before my time @Hero

The one I described as "whack-a-mole" could be the GEBACA, and now I think of the name, it rings a (baby) bell.

I think that some of these were shipped as part of "exptools", which was side-of-desk projects from various people at AT&T. I presume that like me, you worked for one of the AT&T/Bell companies.

The 630s had a 1024x1024 square screen, that gave more space. Because I was only doing local site support, which was a grudging acceptance that I needed access to one in order to be able to offer help to other users, I had to settle for an amber screen one, with what I believe was called the "commercial" 122 key keyboard, that had two rows of function keys above the numbers, like the IBM 3270 data terminals.

In Europe, the distribution and maintenance of these terminals was through Olivetti, as AT&T did not have a channel for selling this type of kit in Europe.

[John Smith 19]

Anyone thinking about Rob Pike's "Help" system?

Not really "graphical" but everything-is-clickable aradign

His paper is quite fascinating (if you're into UI design that is) on what's the bare, irriducible minimum you need, and how to expose program functions (IE stuff you pick off a dropdown menu) to other software.

But just a really different take on what's needed, and what's cruft.

Sadly I don't think it went much further, which is a pity.

[longtimeReader]

CDE seems a bit modern for me

I never liked the CDE environment on my AIX systems - in fact I'm still running Motif/mwm there. The only real difficulty I've had with it was when I needed to make some of the fonts available via my Linux machine which has one of its monitors used as a sort of XTerminal.

Just last week I cleaned out my home directory of the various config files and directories that had been left for so many years by some of the other desktop environments I'd tried and discarded.

[chasil]

OpenBSD fvwm

You will be very pleased with OpenBSD's default window manager.

It's not quite mwm, but it's very close.

[Morten Bjoernsvik]

4dwm

Anyone knows where to get a 4dwm for Ubuntu distros. I especially loved how it had this window of running apps. Or you could stach each running app as icons when minimized and no launcher. I had 4 1280*1024 heads on my indigo2 maximum impact back in 1997.

[Anonymous Coward]

Re: 4dwm

AFAIK 4Dwm was never ported anywhere beyond the native SGI IRIX.

However, websearch claims there are a couple lookalikes, eg. 5Dwm, and 4Dwm themes for Fvwm, IceWM et al. I'm familiar with none of them and haven't touched an IRIX system in far too long (more's the pity), but now I'm curious.

I'd love to see a 4Dwm theme for Xfce -- much easier for me to try that out these days.

[karlkarl]

There is another CDE contender. Here is my old attempt back around 2012:

https://web.archive.org/web/20120602004147/http://devio.us/~kpedersen/screenshots.php

You can still find a few forks on GitHub.

It was mildly popular with those wanting a standard desktop experience. I only stopped because I was asked by the team releasing the original CDE open-source to help :)

[jonnycando]

I so must install this one!

[Mint Sauce]

Jumpers for goalposts

Man, those images bring back some memories. Then again I remember thinking that Suntools was pretty amazing back in the day ;-)

[eddieddieddie]

But on a touchscreen?

I bet this would not work well with a touch screen interface?

[Sin2x]

I prefer my fonts jagged and crispy clear, thank you very much. Leave the blurriness to the applophiles.

[jonnycando]

Well foo! I cloned the GIT repo and attempted to compile....error thrown account missing header files....like how am I supposed to know what's missing?...I cloned the whole repo and everything in it!

[Ozan]

I am now very much inquired to do slack builds for it.

[jake]

https://slackbuilds.org/repository/15.0/desktop/NsCDE/?search=NsCDE